Therapeutic potential of self-antigen-specific CD4+CD25+ regulatory T cells selected in vitro from a polyclonal repertoire

Authors


Abstract

CD4+CD25+ regulatory T cells (Treg) play a major role in the prevention of autoimmune diseases. Converging evidence indicates that Treg specific for self-antigens expressed by target tissues have a greater therapeutic potential than polyclonal Treg. Therefore, the selective expansion of rare self-antigen-specific Treg naturally present in a polyclonal repertoire of Treg is of major importance. In this work, we investigated the potential of different dendritic cell (DC) subsets to expand antigen-specific Treg in mice. The in vitro selective expansion of rare islet-specific Treg from polyclonal Treg could only be achieved efficiently by stimulation with CD8+ splenic DC presenting islet antigens. These islet-specific Treg exerted potent bystander suppression on diabetogenic T cells and prevented type 1 diabetes. This approach opens new perspectives for cell therapy of autoimmune diseases.

See accompanying commentary: http://dx.doi.org/10.1002/eji.200635967

Abbreviations:
GITR:

glucocorticoid-induced TNFR family-related receptor

HA:

hemagglutinin

l-DC:

(lymphoid) CD11chighCD8+CD11b cells

m-DC:

(myeloid) CD11chighCD8CD11b+ cells

NK-DC:

CD11cintB220DX5+ cells

p-DC:

plasmacytoid CD11cintB220+DX5 cells

T1D:

type 1 diabetes

Treg:

CD4+CD25+ regulatory T cells

Introduction

The existence of suppressor cells involved in the regulation of T cell functions has been suggested for more than three decades. It is only in the nineties that identification of some of them has been narrowed down among the CD25-expressing CD4+ T cells 1. Since then, CD4+CD25+ natural regulatory T cells (Treg) have been progressively considered as cells playing a major role in peripheral tolerance to auto-antigens 2. Several reports also describe quantitative or functional deficiency of Treg in different autoimmune diseases in human 3.

Recent data in mice showed that pancreatic islet-specific Treg efficiently suppress type 1 diabetes (T1D), whereas polyclonal Treg are poorly active 48. These reports and others 9, 10 suggest that potent and specific tolerance to nominal Ag can be induced by transferring Treg specific for these Ag. It is now essential to design conditions to expand, from a normal repertoire of polyclonal Treg, the rare cells specific for Ag expressed by the target tissue.

In this work, we show that high background proliferation was observed when Treg were stimulated by autologous DC and IL-2, rendering difficult the selection of rare specific Treg. This obstacle was overcome by using specifically the CD8+ splenic DC to stimulate the Treg. With CD8+ DC as APC, we were able to selectively expand, from a polyclonal Treg repertoire, islet-specific Treg, which efficiently suppress T1D.

Results

CD8+ splenic DC are optimal APC for selecting Ag-specific Treg

Recent data showed the therapeutic potential of Treg specific for islet Ag 4, 5, 7. Our goal is now to define culture conditions allowing the selective expansion of Treg specific for defined rare self-Ag among the natural and diverse repertoire of polyclonal Treg from un-manipulated mice. This process necessitates conditions where the basal proliferation background obtained in culture without added Ag is very low, favoring preferential expansion of specific Treg when an exogenous Ag is added. In a first step, we compared various APC for their capacity to stimulate specific Treg purified from TCR-transgenic mice. We used Treg derived from TCR-HA126–138 mice that were specific for the peptide 126–138 of the hemagglutinin (HA) of Influenza virus presented by I-Ad. The cells were purified on the basis of their CD4+CD25highCD62Lhigh phenotype to limit the risk of possible contamination with activated conventional CD4+ T cells that have down-regulated CD62L expression. As compared to the splenic CD11c fraction containing B cells and macrophages as APC, splenic DC were far more efficient to stimulate HA-specific Treg in the presence of the cognate peptide, confirming that DC are potent APC for stimulating Treg 11, 12. However, a proliferation background was consistently obtained in the Treg culture without exogenous peptide, leading to a stimulation index (defined as the proliferation ratio between the culture with and without the Ag) of less than 20 (Fig. 1A). A similar high basal proliferation was also detected in the culture of Treg purified from non-transgenic BALB/c mice and stimulated by autologous splenic DC (not shown). In contrast, very low proliferation background was observed when HA-specific CD25 T cells were stimulated by splenic DC, leading to a high stimulation index of over 100 when the HA peptide was added (Fig. 1A).

Figure 1.

CD8+ splenic DC are good candidate APC for selecting specific Treg. (A) Treg or CD25 cells, purified from TCR-HA126–138-transgenic mice, were stimulated by DC-depleted splenocytes (CD11c) or whole splenic DC (DC) with or without the HA126–138 peptide. IL-2 was added in the Treg culture. After 5 days, cell proliferation was assessed by thymidine incorporation. Mean proliferation (+ SD) of each culture (left panel) and stimulation index (right panel), calculated by dividing the mean cell proliferation in the culture with HA peptide by the mean cell proliferation in the culture without peptide, of one representative experiment, out of three, is shown. (B, C) Treg, purified from TCR-HA126–138 mice, were stimulated by the indicated DC subsets in the presence of IL-2 and HA126–138 peptide (HA peptide), islet lysates from ins-HA mice (β cells) or without exogenous Ag (medium). After 5 days, cell proliferation was analyzed by thymidine incorporation. (B) Mean proliferation (+ SD) of each culture (left panel) and stimulatory index calculated as in A (right panel) of one representative, out of four independent experiments, is shown. (C) Relative average of stimulatory index of the four independent experiments was calculated as followed: For each experiment, the stimulatory index of each culture, calculated as in (A), was divided by the stimulatory index obtained in the culture with l-DC, giving relative values. The average of these values from the four experiments was represented in arbitrary units, the one of the l-DC culture is by definition 1. Error bars represent the SD. Compared to the l-DC culture, relative stimulation index of the three other cultures was significantly lower (p <0.001). (D) Treg purified from non-transgenic BALB/c mice were stimulated by l-DC or m-DC and IL-2. After 5 days, cell proliferation was assessed by thymidine incorporation. Mean proliferation (+ SD) of one representative experiment out of two is shown. Compared to the m-DC culture, proliferation of the l-DC culture was significantly lower (p <0.005).

The high proliferation background observed when Treg were cultured with splenic DC would likely hamper the selection of rare specific Treg. We thus tested the capacity of different DC subsets to stimulate Treg. In mice, at least four DC subsets have been described in the spleen. The CD11chighCD8+CD11b cells, subsequently referred to as l-DC because of their lymphoid related phenotype 13, the CD11chighCD8CD11b+ cells, subsequently referred to as m-DC because of their myeloid related phenotype 13, the plasmacytoid CD11cintB220+DX5 cells, subsequently referred to as p-DC 14, and the CD11cintB220DX5+ cells, subsequently referred to as NK-DC because of their expression of NK markers 15. HA-specific Treg were stimulated by each of these four DC subsets with or without HA peptide. In the presence of the peptide, m-DC induced very potent Treg proliferation, as well as l-DC, although with a reduced capacity. Moderate and poor Treg proliferation was obtained, respectively, with p-DC and NK-DC. Importantly, a high proliferation background (culture without HA peptide) was obtained when Treg were stimulated by m-DC but not by l-DC, p-DC or NK-DC. Consequently, l-DC was the only DC subset allowing a high proliferation of specific Treg in the presence of the cognate Ag and low proliferation background without peptide. This was measured by the stimulation index, which was 55 for l-DC and below 15 for the three other DC subsets (Fig. 1B and C). We also observed that Treg purified from BALB/c non-transgenic mice proliferated spontaneously at high level to autologous m-DC but not to autologous l-DC (Fig. 1D). Thus, the low proliferation background observed when Treg were stimulated by l-DC did not rely on some unique characteristic of TCR-transgenic cells.

For a given autoimmune disease, the panel of tissue-specific peptides that could be used to select Treg that have therapeutic effects would be different for each patient and sometimes would even not be known. It was thus essential to verify that HA-specific Treg can be also stimulated by whole lysate of islet cells expressing HA, instead of adding purified HA peptide. Lysate of HA+ islet cells was prepared from ins-HA mice expressing HA under the control of the insulin promoter. When pulsed with this cell lysate, both m-DC and l-DC induced a significant proliferation of specific Treg, although reduced when compare to culture with HA peptide, whereas p-DC or NK-DC were poor APC. Again, the stimulation index for l-DC was significantly higher than the one for the three other DC subsets (Fig. 1B and C). In these assays, Treg proliferation was effectively due to HA expressed by the β-cell lysates from ins-HA mice. Indeed, low background Treg proliferation (<1000 cpm) was observed when DC subsets were pulsed with β-cell lysates from BALB/c mice (data not shown). These experiments suggest that l-DC are the best APC for selecting rare Ag-specific Treg.

High expansion of functional Ag-specific Treg stimulated by l-DC

Generating enough Ag-specific Treg for a therapeutic application will require the capacity not only to select these rare cells but also to expand them. We thus tested the capacity of l-DC to expand Ag-specific Treg. Treg from TCR-HA126–138 mice were stimulated by HA126–138-pulsed l-DC for the first 2 weeks of the selection phase and then by HA126–138-pulsed splenic DC. During the first 3 weeks of culture, 200-fold expansion of Treg was observed (Fig. 2). Interestingly, over 3 weeks of culture, similar expansion was found when HA-specific Treg were stimulated by HA-pulsed m-DC and HA-pulsed l-DC (not shown). Thus, whereas m-DC had superior stimulatory capacity than l-DC in short-term culture (Fig. 1), both DC subsets were equally good stimulators of Treg in long-term culture. In the absence of the HA peptide, Treg stimulated by l-DC progressively died in the culture (Fig. 2). Treg expanded by HA-pulsed l-DC exhibited a phenotype of natural suppressor cells since they expressed high levels of FoxP3 (Fig. 3A), CD4, CD25 and glucocorticoid-induced TNFR family-related receptor (GITR) (Fig. 3B). Although extensively activated, they also maintained CD62L expression (Fig. 3B), as previously reported in another type of Treg culture 16.

Figure 2.

High expansion of Ag-specific Treg stimulated by l-DC. Treg, from TCR-HA126–138-transgenic mice, were stimulated by l-DC at days 0 and 7, followed by whole DC at day 14, and IL-2 with or without HA126–138 peptide. Mean values (+ SD) of live cell expansion, excluding trypan blue, obtained from three independent experiments, are shown.

Figure 3.

Natural Treg phenotype of expanded Treg. (A) Quantitative real time RT-PCR for FoxP3. Total RNA was extracted from fresh cells (CD25 and CD25+ cells from spleen and lymph nodes or whole splenic DC), from expanded Treg originated from TCR-HA126–138 mice (after culture with the HA126–138 peptide obtained as in Fig. 2), or from expanded Treg originated from ins-HA transgenic mice (after culture with HA peptides or anti-CD3 obtained as in Fig. 6). Levels of FoxP3 mRNA were determined by quantitative real time RT-PCR analysis in triplicates (mean ± SD). (B) Expanded Treg generated as described in Fig. 2 from TCR-HA126–138 mice were labeled with the indicated markers as described in Materials and methods section.

Expanded Treg exhibited in vitro suppressive activity since they efficiently inhibited proliferation of HA126–138 specific CD25 conventional T cells stimulated by HA126–138 pulsed DC (Fig. 4A). While freshly purified Treg need to be activated via their TCR to suppress 17, 18, expanded Treg require 5 or not 17 TCR engagement to turn-on their suppressive activity, depending on the duration and the conditions used for their expansion. We thus analyzed whether expanded Treg generated in our conditions exhibited constitutive suppressive activity. Expanded HA126–138-specific Treg efficiently inhibited proliferation of HA111–119-specific CD25 conventional T cells only in the presence of DC pulsed with both HA126–138 and HA111–119 peptides. When cells were stimulated by DC pulsed only with the HA111–119 peptide, minimal Treg suppression was observed, showing that these expanded Treg needed to be re-stimulated via their TCR to suppress (Fig. 4A).

Figure 4.

Expanded Ag-specific Treg stimulated by l-DC have suppressive activity. (A) After 3 weeks of culture with the HA126–138 peptide, expanded Treg were analyzed for their in vitro suppressive activity. CD25 responder T cells, purified from TCR-HA126–138 or TCR-HA111–119 mice, were cocultured with expanded HA126–138-specific Treg at the indicated ratio, in the presence of splenic DC pulsed with HA126–138 and/or HA111–119 peptides as indicated. After 96 h of culture, Mean cell proliferation (+ SD) was assessed by thymidine incorporation. One representative experiment of two is shown. (B) Unexpanded HA-specific Treg freshly purified from TCR-HA111–119 were analyzed for their capacity to prevent diabetes induced in ins-HA-transgenic mice. In control animals injected only with diabetogenic CD25 cells (1.5 × 106), mice developed T1D within 6–10 days (line without symbol, n = 2). Co-transfer of high dose (2 × 106, solid circles, n = 3) or low dose (0.5 × 106, open circle, n = 3) of unexpanded HA-specific Treg completely or partially prevented from diabetes development, respectively. The graph shows evolution of the incidence of diabetes. One representative experiment of two is shown. (C) After 2 or 3 weeks of culture, expanded Treg derived from TCR-HA126–138 mice and stimulated by HA-pulsed l-DC (expanded HA-specific Treg) or polyclonal expanded Treg derived from BALB/c mice and stimulated by l-DC in the presence of anti-CD3 mAb (expanded polyclonal Treg) were analyzed for their capacity to prevent diabetes induced in ins-HA-transgenic mice. In control animals injected only with diabetogenic CD25 cells (1.5 × 106), mice developed T1D within 6–10 days (line without symbol, n = 12). Co-transfer of high dose (4 × 106, solid circles, n = 10) or low dose (0.4 × 106, open circle, n = 5) of expanded HA-specific Treg completely or partially prevented from diabetes development, respectively. In contrast, high dose (4 × 106) of expanded polyclonal Treg had no effect (dashed line, n = 8). The graph shows evolution of the incidence of diabetes. Data were pooled from two independent experiments, excepted for the low dose of HA-specific Treg (one experiment).

Expanded Treg were also tested for their in vivo suppressive activity. We thus designed a new mouse model of T1D in ins-HA mice expressing HA under the control of the insulin promoter conferring an islet β cell restricted expression. Ins-HA mice were sub-lethally irradiated and then transferred with CD25-depleted HA-specific T cells derived from TCR-HA111–119 mice, consequently enriched in T cells specific for the HA111–119 peptide presented by I-Ed. Then, mice were immunized with HA-pulsed DC to activate the diabetogenic T cells, inducing T1D within 6–10 days in all the animals (Fig. 4B) and severe insulitis (not shown). We first checked whether unexpanded HA-specific Treg freshly purified from TCR-HA111–119 mice were able to suppress diabetes induced in ins-HA mice. A dose-response effect was observed with complete prevention of T1D in mice receiving high dose of HA-specific Treg (Fig. 4B). To test the in vivo suppressive activity of expanded HA-specific Treg, ins-HA mice were co-transferred with conventional HA111–119-specific T cells and high number of expanded HA126–138-specific Treg. None of these mice developed clinical diabetes within the 35 days of the experiment. A partial protection was observed when tenfold less expanded HA-specific Treg were injected (Fig. 4C). In contrast, injection of the same number of polyclonal expanded Treg derived from BALB/c mice did not regulate the disease confirming that only islet-specific Treg efficiently suppressed T1D (Fig. 4C). Thus, expanded HA-specific Treg were functional to block T1D in ins-HA mice, even when Treg and diabetogenic T cells recognized different peptides presented respectively by I-Ad and I-Ed, showing bystander suppression exerted by Tregin vivo.

Selective expansion of rare Ag-specific Treg stimulated by l-DC

We then evaluated whether culture conditions described above allowed the selection of rare Ag-specific Treg. We first checked that l-DC were superior to m-DC for this selection. HA126–138 specific Treg, expressing the Thy-1.1 congenic marker, were diluted into non-transgenic Thy-1.2 BALB/c Treg at a 1 to 1000 ratio. After 3 weeks of culture, the proportion of HA126–138-specific Treg was increased by a factor 80 when stimulated by l-DC but only by a factor 10 when stimulated by m-DC in the presence of the cognate HA126–138 peptide (not shown). Then, to have a more accurate evaluation of the potential of l-DC to select rare specific Treg, we diluted down HA111–119-specific Treg, identified with the 6.5 anti-clonotypic Ab, in BALB/c Treg. We started with four cultures in which the proportions of HA111–119-specific Treg represented 10%, 1%, 0.1% or 0.01% of the cells, subsequently referred to as respectively “1/10 HA”, “1/100 HA”, “1/1,000 HA” and “1/10,000 HA” Treg cultures. Cells were stimulated by l-DC pulsed with the HA111–119 peptide. After 4 weeks of culture, expanded cells exhibited a Treg phenotype with high-level expression of CD4 and CD25 (not shown). Among them, the proportion of HA-specific 6.5+ Treg was determined by flow cytometry. Between day 0 and day 28 of the cultures, the proportions of HA-specific Treg rose from 10 to 83%, 1 to 62%, 0.1 to 35% and 0.01 to 5% (Fig. 5A). During the same period, dramatic expansion of HA-specific Treg was also observed. A 1200-, 2900-, 6000- and 11 000-fold expansion of HA-specific Treg was calculated for respectively the “1/10 HA”, “1/100 HA”, “1/1,000 HA” and “1/10,000 HA” Treg cultures (Fig. 5A). The selection and expansion process of rare Treg (1 specific Treg out of 10 000 at day 0 of the culture) was particularly efficient since the proportion of HA-specific Treg was increased by 500-fold and their numbers were increased by 11 000-fold.

Figure 5.

Selection of rare Ag-specific Treg stimulated by l-DC. HA111–119-specific Treg purified from TCR-HA111–119 transgenic mice were diluted in the polyclonal population of Treg purified from non-transgenic BALB/c mice at different ratios. At day 0, the proportions of HA111–119-specific Treg, identified by the 6.5 anti-clonotypic Ab, represented 10% (1/10 HA), 1% (1/100 HA), 0.1% (1/1,000 HA) or 0.01% (1/10,000 HA) of the cells. The four cultures were stimulated by l-DC at days 0 and 7, followed by whole DC at days 14 and 21 in the presence of HA111–119 peptide and IL-2. Cells were re-stimulated by HA-pulsed l-DC at day 7 and by HA-pulsed CD11c+ splenic DC at days 14 and 21 of the culture. (A) The proportions of 6.5+ cells were analyzed by flow cytometry to determine the increase of the percentages (left panel) and the absolute numbers (right panel) of HA-specific Treg between day 0 (d0) and day 28 (d28) for the four cultures. (B) The same expanded Treg were analyzed for their capacity to prevent diabetes induced in ins-HA-transgenic mice. Mice were injected with diabetogenic CD25 T cells (1.5 × 106) alone (No Treg) or co-transferred with expanded cells (2 × 106) from the 1/10 HA, 1/100 HA, 1/1000 HA or 1/10 000 HA Treg cultures. The graphs show the evolution of blood glucose levels. Each dot represents an individual mouse and the bars represent the mean of glucose levels from three to five mice per group. The vertical dotted lines represent the time when mice of the control group (no Treg) became hyperglycemic.

These expanded Treg were also tested for their capacity to prevent the development of diabetes in ins-HA transgenic mice. The disease was obtained as described above. Controls with no Treg became diabetic 7–8 days after transfer of effector T cells. Mice receiving the “1/10 HA” or the “1/100 HA” Treg were fully protected from clinical diabetes. Mice receiving the “1/1,000 HA” Treg were partially protected as two mice exhibited a delayed hyperglycemia (over 300 mg/dL), which progressively returned to lower blood glucose levels and a third mouse did not develop clinical diabetes (glucose levels below 200 mg/dL). Injection of “1/10,000 HA” Treg had no effect (Fig. 5B). Thus, in this setting, injection of Treg protected from diabetes only when a significant fraction of its repertoire was specific for islet Ag.

Inhibition of T1D by islet-specific Treg selected from a diverse repertoire

Experiments described in Fig. 5 show that rare Treg can be selected and expanded, and, after expansion, have in vivo functional activity upon transfer. This suggests that rare Treg specific for pancreatic islet-Ag may be selected from a diverse polyclonal repertoire using our culture conditions. To test this hypothesis, we explored whether HA-specific Treg can be selected from the Treg repertoire of ins-HA mice in which HA is a neo-self-Ag with peripheral expression restricted to pancreatic islet cells 19. In this culture, referred to as “l-DC HA-Treg”, Treg were stimulated by l-DC pulsed with both HA111–119 and HA126–138 peptides. As a control, referred to as “no-Ag-Treg”, Treg were stimulated by non-pulsed l-DC. As an additional control, referred to as “aCD3-Treg”, Treg were stimulated by anti-CD3 Ab-pulsed l-DC in order to expand the whole repertoire of Treg. Within 4 weeks of culture, the absolute number of l-DC HA-Treg was increased by a factor 12 and the one of aCD3-Treg by a factor 80. Cells of the three cultures, analyzed after 3 weeks, had a natural Treg phenotype with high-level expression of Foxp3 (Fig. 3A), CD4, CD25, CD62L, and GITR (Fig. 6A).

Figure 6.

Inhibition of T1D by islet Ag-specific Treg selected from a diverse repertoire. Treg, purified from ins-HA-transgenic mice, were stimulated by l-DC or m-DC at days 0 and 7, followed by whole DC and splenocytes at days 14 and 21 in the presence of IL-2 and either HA111–119 + HA126–138 peptides (l-DC HA-Treg and m-DC HA-Treg) or anti-CD3 Ab (aCD3-Treg) or without exogenous Ag (No-Ag-Treg). After 3 weeks of culture, expanded l-DC HA Treg were labeled with the indicated markers as described in Materials and methods section (A) and were analyzed for their in vitro suppressive activity (B). HA126–138-specific CD25 responder cells were cocultured with expanded Treg at the responder/Treg indicated ratios in the presence of DC pulsed with HA111–119 + HA126–138 peptides (HA peptides) or anti-CD3 Ab-pulsed DC (CD3 mAb). After 96 h of culture, cell proliferation was assessed by thymidine incorporation. (C) After 4 weeks of culture, aCD3-Treg, l-DC HA-Treg and m-DC HA-Treg were analyzed for their capacity to prevent diabetes induced in ins-HA mice. No-Ag-Treg that poorly expanded could not have been tested in vivo. Mice were injected with diabetogenic CD25 T cells (0.75 × 106) alone (No Treg, n = 4), or were co-transferred with aCD3-Treg (3 × 106, x, n = 6), or l-DC HA-Treg (3 × 106, n = 4), or m-DC HA-Treg (3 × 106, n = 4). The graph shows the evolution of the incidence of diabetes.

To test the specificity for HA peptides of expanded Treg, cells were cocultured with HA-specific CD25 responder T cells in the presence of HA-pulsed DC. As, in our culture conditions, expanded Treg became suppressive only after re-activation (Fig. 4A), inhibition of responder T cell proliferation would indicate that a significant fraction of Treg have been re-activated by HA-pulsed DC and were thus HA-specific. l-DC HA-Treg suppressed efficiently T cell proliferation even at a ratio of one Treg for two responder T cells. In contrast, aCD3-Treg and no-Ag-Treg did not exhibit any suppressive activity in this HA-specific T cell proliferation assay (Fig. 6B). As a control to test the functionality of Treg from the three cultures, cells were stimulated by anti-CD3 Ab-pulsed DC, which turn on suppressive activity of Treg whatever their Ag specificity 18, 20. Treg from the three different cultures exhibited similar strong suppressive activity (Fig. 6B). These in vitro experiments showed that l-DC HA-Treg were indeed enriched in HA-specific cells, contrary to aCD3-Treg or no-Ag-Treg controls.

We then analyzed the capacity of the expanded Treg to regulate T1D in the ins-HA model. Mice transferred with the HA-specific CD25 T cells alone developed clinical diabetes 6 to 13 days after cell transfer. The co-transfer of nonspecific aCD3-Treg had no effect since similar kinetics of diabetes onset was observed. In contrast, injection of l-DC HA-Treg prevented diabetes in the ins-HA model. Except for one mouse that had a delayed diabetes onset (day 18), other mice remained normo-glycemic for the duration of the experiment (Fig. 6C). These experiments show that rare islet-specific Treg can be selectively expanded in vitro up to a level where they can be used to prevent T1D.

To further confirm that l-DC were more appropriate than m-DC to select rare Ag-specific Treg, we also stimulated purified Treg from ins-HA transgenic mice by m-DC pulsed with both HA peptides, generating “m-DC HA-Treg” expanded cells. While both l-DC HA-Treg and m-DC HA-Treg were highly suppressive when re-stimulated by anti-CD3 Ab, m-DC HA-Treg had reduced suppressive activity upon re-stimulation with HA-pulsed DC, compared to l-DC HA-Treg. This correlated with a poor capacity of m-DC HA-Treg to control diabetes, contrary to l-DC HA-Treg (Fig. 6).

Discussion

A major challenge in improving the treatment of immunopathologies such as autoimmune diseases will consist of inducing a state of long-term specific tolerance in order to block the preferential activation of pathogenic T cells without affecting the rest of the immune system. New treatments are being proposed to reach this goal. CD3-specific Ab administration is a promising approach that has recently been proposed in new diabetic onset patients 21, 22. Vaccination by administration of target Ag has been shown to efficiently regulate autoimmune diseases in mouse models 23. Cell therapy using Treg specific for auto-Ag expressed by target tissue is another attractive strategy 47. However, this latter approach has been obtained with Ag-specific Treg purified from TCR-transgenic mice. For a possible therapeutic application, it is essential to show that Ag-specific Treg can be selected from a diverse repertoire of polyclonal Treg.

A recent report showed that islet-Ag-specific Treg could be selectively expanded from the polyclonal repertoire of NOD Treg using beads coated with recombinant islet peptide mimic-MHC class II complex and anti-CD28 mAb 24. Here, we showed that such a selection process can be achieved using an alternative and complementary strategy. Indeed, using l-DC to select specific Treg present a major and unique advantage. These cells have been shown to be very efficient in cross-presenting proteins or cell lysates 2527. In accordance with this property, we found that HA-specific Treg were efficiently activated by l-DC pulsed with lysates of beta cells expressing the cognate HA protein. This technology would be particularly adapted to expand Treg with therapeutic effects. Indeed, in autoimmune diseases, the peptide specificities of Treg that would have a therapeutic potential would depend on known (MHC haplotype, type of autoimmune disease) and unknown or unpredictable factors (frequency of specific Treg in their repertoire, efficiency of these specific Treg in this disease). Choosing the appropriate selecting peptides would thus be difficult, if not impossible. As an alternative to the culture with purified tissue-specific peptides, Treg could be stimulated by cell lysates derived from the target tissue of the autoimmune process, generating tissue-specific Treg that have therapeutic activity in this particular disease.

Others 11, 28 and we experienced background proliferation when Treg were stimulated by whole splenic DC or bone marrow-derived DC that complicates the selection of rare Ag-specific Treg. In this work, we observed that such proliferation background, which is likely due to the autoreactivity of the Treg repertoire 29, 30, was not observed when using the l-DC as APC. A possible explanation could be related to the fact that cells of hematopoietic origin, likely the thymic DC, are involved in negative selection of highly autoreactive Treg in the thymus 3133. Since thymic DC and l-DC have related features 13, Treg recognizing the panel of endogenous peptides presented by l-DC would be less represented in the periphery because of their thymic deletion, giving low proliferation background when Treg were stimulated by l-DC. On the contrary, some of the self-peptides constitutively presented by m-DC, but not by l-DC, would be specifically recognized by a fraction of autoreactive Treg not deleted in the thymus. Cells of this fraction would be the ones that would be preferentially stimulated by m-DC.

During the culture to selectively expand islet-specific Treg, cells maintained high levels of CD62L expression, which could be an important feature for a therapeutic use. Indeed, Treg that have therapeutic effects in T1D are contained within the fraction expressing high levels of CD62L 34, 35, probably because they act after migration in pancreatic lymph nodes. In addition, others and we have previously reported that islet-specific Treg, but not nonspecific Treg, are preferentially activated in pancreatic lymph node 36, 37. Consequently, islet-specific Treg would have the unique capacity to induce a localized suppression that act on diabetogenic T cells. Additionally, we also showed that Treg specific for an islet peptide could regulate the disease induced by diabetogenic T cells specific for another islet peptide. This bystander or linked suppression is a critical parameter because it suggests that injecting Treg specific for one particular islet Ag would regulate T1D resulting from the activation of T cells specific for multiple islet Ag, as previously discussed 4, 5. The specificity of action of Treg thus relies on the location of their sustained activation in vivo. For instance, islet-specific Treg that are preferentially activated in the pancreas or its draining lymph nodes, would be efficient to regulate T1D or any immune response in this region but would not affect the rest of the immune system. In this line, we observed that ins-HA mice, protected from development of diabetes by injection of HA-specific Treg, were immuno-competent. Indeed, they were able to reject skin allografts within 8–12 days, a kinetics similar to the one observed in control mice (not shown).

Further experiments should be performed in order to show that functional islet-specific Treg could be expanded when stimulated by l-DC pulsed with β-cell lysates in the more physiological diabetes model of NOD mice. The same approach could be used also to select Treg specific for allo-Ag or allergens in order to induce specific tolerance in transplantation or allergy. Additional experiments will be required to adapt culture conditions for selecting Ag-specific human Treg.

Materials and methods

Mice

Six- to eight-week-old female BALB/c mice were obtained from Charles River Laboratories (l'Arbresle, France). The ins-HA-transgenic mice expressing HA of Influenza virus in pancreatic islet β cells 19 were backcrossed more than ten generations onto the BALB/c genetic background. The TCR-HA126–13838 and TCR-HA111–11939 transgenic mice expressing a TCR recognizing respectively HA 126–138 epitope presented by I-Ad and HA 111–119 epitope presented by I-Ed were backcrossed more than ten generations onto the BALB/c genetic background. All mice were bred in our specific pathogen-free animal facility. They were manipulated according to European Union guidelines.

Antibodies and flow cytometric analyses ofexpanded Treg

After harvesting cultivated cells, cells were pre-incubated with 2.4G2 mAb (PharMingen) to block nonspecific binding to Fc receptors and then stained in PBS 3% FCS with saturating amounts of combinations of the following mAb: FITC-conjugated anti-CD4 (GK1.5, PharMingen), phycoerythrin-conjugated anti-CD25 (PC61, PharMingen), biotin-conjugated anti-CD62L (MEL-14, PharMingen), biotin-conjugated anti-GITR (polyclonal goat Ab, R&D systems) or biotin-conjugated isotypic Ab controls (PharMingen). The biotinylated mAb were detected by Cy-Chrome streptavidin (PharMingen). FACScalibur analyses were performed with CellQuest software (Becton Dickinson).

Purification of CD4+CD25highCD62Lhigh Treg and CD25 cells

After a mechanical dissociation, cells from spleen and peripheral lymph nodes were incubated with biotin-conjugated anti-CD25 mAb (7D4, PharMingen), streptavidin microbeads (Miltenyi Biotec), followed by two consecutive magnetic cell separations using LS columns (Miltenyi Biotec). The CD25 cells, which did not bind to the beads, were harvested from the flow through and contained less than 0.3% of CD4+CD25+ T cells. Then, Treg were further enriched by flow cytometry. Cells were stained with FITC-conjugated anti-CD4 (GK1.5, PharMingen), phycoerythrin-conjugated anti-CD62L (MEL-14, PharMingen) and streptavidin-Cy-Chrome (PharMingen) which bound to free biotin-conjugated anti-CD25 Ab, uncoupled to beads. The CD4+CD25highCD62Lhigh T-cells were sorted on a FACstar+ (Becton Dickinson), giving a purity up to 99%.

APC purification

Spleens from BALB/c mice were digested with liberase (1.67 Wünsch U/mL, Boerhinger Mannheim) and DNase (0.1 mg/mL, Boerhinger Mannheim) diluted in RPMI (Gibco BRL). Cells were then incubated with anti-CD11c-coated microbeads (Miltenyi Biotec), followed by two2 consecutive magnetic cell separations using LS columns (Miltenyi Biotec), giving a purity of 95%. The CD11c-depleted splenocytes were harvested from the flow through. All steps were performed in PBS with 3% serum and 2 mM EDTA. To purify the lymphoid (l-DC: CD11chighCD8+CD11b) and myeloid (m-DC: CD11chighCD8CD11b+) DC subsets, cells were stained with FITC-labeled anti-CD11b (M1/70, PharMingen), phycoerythrin-labeled anti-CD11c (HL3, PharMingen) and Cy-Chrome-labeled CD8a (53–6.7, PharMingen). To purify the plasmacytoid DC (p-DC: CD11cintB220+DX5) and NK-DC (CD11cintB220DX5+), cells were stained with FITC-labeled anti-B220 (RA3–6B2, PharMingen), phycoerythrin-labeled anti-CD11c (HL3, PharMingen) and biotin-labeled anti-DX5 NK-cell marker (DX5, PharMingen). The different DC subsets were sorted on a FACstar+ (Becton Dickinson), giving a purity of at least 98.5%.

Islet lysates

Pancreatic islets, isolated from ins-HA-transgenic mice as described 40, were digested with 1.4 mg/mL dispase (Roche) diluted in PBS with 5% bovine albumin (Sigma). Cells were lysed by freeze and thaw cycles.

T cell proliferation assay

CD4+CD25highCD62Lhigh Treg or CD25 cells (2 × 104) were stimulated by 3 × 103 irradiated (25 Gy) CD11c-depleted splenocytes or whole splenic CD11c+ DC or DC subsets pulsed with the HA126–138 peptide (10 µg/mL, Neosystem) or 15 × 103 lysed islet cells from ins-HA-transgenic mice in 10% FCS RPMI, and supplemented with murine GM-CSF (10 ng/mL, R&D systems) and, for Treg, murine IL-2 (10 ng/mL, R&D systems) in 96-well flat-bottom plates for 5 days. Cells were pulsed with [3H-methyl]thymidine (Amersham) for the last 15 h.

In vitro Treg expansion

CD4+CD25highCD62Lhigh Treg (105/mL) were stimulated by l-DC (2 × 104/mL) pulsed with 10 µg/mL HA126–138 peptide (Neosystem) and/or 1 µg/mL HA111–119 peptide (Neosystem), or 1 µg/mL anti-CD3 mAb (145–2C11, PharMingen) in 10% FCS RPMI and 10 ng/mL GM-CSF and 10 ng/mL IL-2 (R&D systems). Treg were similarly re-stimulated at day 7 with freshly purified l-DC. At days 14 and 21, Treg were re-stimulated by whole splenic CD11c+ DC (Treg/DC ratio of 5/1) and cytokines (GM-CSF and IL-2), supplemented with whole irradiated splenocytes (Treg/splenocyte ratio of 1/1) for the culture of Treg originated from ins-HA mice. Expanded Treg were harvested 7 days after the last re-stimulation for in vitro and in vivo analyses.

Real-time PCR

Primer and probe sequences for FoxP3 and HPRT were previously described 41 and purchased from Eurogentec. The real-time PCR was performed on an ABI prism 7700 using Taqman Universal PCR master mix (Applied Biosystems) in triplicates. The average threshold cycles (CT) were used to calculate the relative expression ratios (tested samples/fresh CD25 cells) using the comparative CT method as described by Applied Biosystems in the User Bulletin # 2, ABI PRISM 7700 Sequence Detection System. CT for HPRT was used to normalize the samples.

Suppression assays

CD25 cells from TCR-HA126–138 or TCR-HA111–119 (5 × 104) mice were cocultured with different numbers of expanded Treg in the presence of 1 × 104 whole splenic freshly purified DC pulsed with 10 µg/mL HA126–138 peptide and/or 1 µg/mL HA111–119 peptide or 1 µg/mL anti-CD3 mAb in RPMI supplemented with 0.5% mouse serum and 10 ng/mL GM-CSF. Cells were cultured in 96-well round-bottom plates for 96 h and pulsed with [3H-methyl]thymidine (Amersham) for the last 18 h.

Diabetes induction

CD25 cells from TCR-HA111–119 mice were injected intravenously, with or without expanded Treg, into sub-lethally irradiated (3 Gy) ins-HA-transgenic mice. Then, mice were immunized by intravenous injection of splenic DC, matured after overnight culture, and pulsed with the HA126–138 and HA111–119 peptides. Blood glucose levels were monitored using a glucometer (One touch Basic, LifeScan).

Statistical analyses

The statistical significance was determined by the Student's t-test.

Acknowledgements

We are grateful to Harald von Boehmer for providing the TCR-HA transgenic mice. We thank Gilles Marodon for critical reading of the manuscript; Vanessa Dubus Bonet, Benoît Barrou, Sébastien Giraud, Simon Blanchard, Antoine Leclerc and Béatrice Levacher for their contribution to some of the experiments and Gwenaelle Piriou and François Bodin for taking care of our animal colony. This work was supported by the Roche Organ Transplantation Research Foundation, the “Association Française contre les Myopathies”, the Ministère de la Recherche (Action Concertée Incitative Jeunes Chercheurs), the Association Française des Diabétiques, INSERM and Assistance Publique-Hôpitaux de Paris as a Contrat d'Interface (to B.L.S.)

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH

  5. 5

    WILEY-VCH

  6. 6

    WILEY-VCH

Ancillary